Vol. 66, No. 12JOURNAL OF VIROLOGY, Dec. 1992, p. 7146-71520022-538X/92/127146-07$02.00/0Copyright © 1992, American Society for Microbiology

Point Mutations in the DNA Polymerase Gene of HumanCytomegalovirus That Result in Resistance to

Antiviral AgentsNELL S. LURAIN,' KENNETH D. THOMPSON,12* EARLE W. HOLMES,'

AND G. SULLIVAN READ2t

Departments ofPathology' and Microbiology, 2 Loyola UniversityMedical Center, Maywood, Illinois 60153

Received 22 June 1992/Accepted 31 August 1992

Three independently isolated mutants of human cytomegalovirus strain AD169 were found to be resistant toganciclovir at a 50%o effective dose of 200 FM. Phosphorylation of ganciclovir was reduced 10-fold inmutant-infected cells compared with AD169-infected cells. All three mutants were also determined to beresistant to the nucleotide analogs (S)-1-[(3-hydroxy-2-phosphonylmethoxy)propyljadenine (HPMPA) and(S)-1-[(3-hydroxy-2-phosphonylmethoxy)propylJcytosine (HPMPC) and hypersensitive to thymine-l-D-arab-inofuranoside (AraT). Single base changes resulting in amino acid substitutions were demonstrated in thenucleotide sequence of the DNA polymerase gene of each mutant. The polymerase mutation contained in oneof the mutants was transferred to the wild-type AD169 background. Ganciclovir phosphorylation in cellsinfected with the recombinant virus produced by this transfer was found to be equivalent to that ofAD169-infected cells. The ganciclovir resistance of the recombinant was reduced fourfold compared with thatof the parental mutant; however, the recombinant remained resistant to HPMPA and HPMPC andhypersensitive to AraT. The ganciclovir resistance of the mutants therefore appears to result from mutationsin two genes: (i) a kinase which phosphorylates ganciclovir and (ii) the viral DNA polymerase.

Human cytomegalovirus (HCMV) is a member of theherpesvirus family. Primary HCMV infections are generallymild or asymptomatic except for congenital infections,which often have serious sequelae. A hallmark of herpesvi-rus infections is the establishment of latency following aprimary infection. Reactivation of latent infection has in-creasingly become a factor contributing to the morbidity andmortality associated with AIDS and organ transplantation (9,11). Ganciclovir (GCV) and phosphonoformic acid (PFA;foscamet) are presently the only approved drugs for thetreatment of HCMV disease, but isolates resistant to thesedrugs have been reported (2, 10, 18, 30, 31).The viral DNA polymerase is the target of many of the

antiviral agents which have been found to be effectiveinhibitors of HCMV replication. These antiviral agents fallinto two groups. The first group, which includes GCV,acyclovir (ACV), and thymine-1-D-arabinofuranoside (AraT),requires monophosphorylation by a virus-encoded enzyme.Di- and triphosphorylation of GCV monophosphate arebelieved to be carried out by cellular enzymes (2, 23, 29).The triphosphate derivative is the active form which isthe substrate for the viral DNA polymerase. Resistance tothese drugs, therefore, can result from mutations in eitherof two viral genes: (i) the kinase that monophosphorylatesthe drugs (22, 29, 32), or (ii) the DNA polymerase (5, 12,13).The second group of antiviral agents includes the nucleo-

side monophosphate analogs (S)-1-[(3-hydroxy-2-phospho-nylmethoxy)propyl]adenine (HPMPA) and (S)-1-[(3-hy-

* Corresponding author.t Present address: Division of Cell Biology and Biophysics,

School of Life Sciences, University of Missouri, Kansas City, MO64110.

droxy-2-phosphonylmethoxy)propyl]cytosine (HPMPC) andthe pyrophosphate analogs PFA and phosphonoacetic acid(PAA). These agents do not require activation by a virus-encoded product to become substrates for the viral DNApolymerase, and therefore, resistance to these compounds isbelieved to result directly from mutations in the DNApolymerase gene (26, 28).Aside from their clinical relevance, mutations conferring

drug resistance can be important tools for mapping viralgenes and determining the functions of the gene products.An extensive amount of work has been done on the mecha-nisms of antiviral resistance of herpes simplex virus (HSV).Many drug-resistant mutants have been used to study theviral DNA polymerase and thymidine kinase genes of HSV,and most of the mutations have been mapped to specificnucleotides (5, 8, 12, 13, 16, 21). In contrast to HSV, muchless is known about mechanisms of resistance in HCMV.The drug-resistant mutants of HCMV that have been re-ported thus far appear to have mutations in the DNApolymerase gene (7, 31) or in the viral kinase (2, 30).Evidence for altered viral functions has been based primarilyon the pattern of resistance to antiviral agents and reducedintracellular drug anabolism.We have isolated three GCV-resistant mutants of HCMV

strain AD169 which are also resistant to HPMPC andHPMPA. We have characterized these mutants by theirsusceptibilities to other antiviral agents, their intracellularGCV anabolic activities, and the nucleotide sequences oftheir DNA polymerase genes.

(Parts of this work were presented in preliminary form atthe XVth International Herpesvirus Workshop, Washington,D.C., 1990, and the 31st Interscience Conference on Antimi-crobial Agents and Chemotherapy, Chicago, Ill., 1991.)

7146

on August 18, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

POLYMERASE MUTANTS OF HUMAN CYTOMEGALOVIRUS 7147

MATERIALS AND METHODS

Cells and virus. Human foreskin fibroblasts (HFF) atpassage 5 to 7 were purchased from Bartels, Bellevue,Wash., and propagated in Eagle's minimum essential me-dium (MEM) supplemented with 10% fetal calf serum, 50 ,ugof gentamicin per ml, and 2.5 ,ug of amphotericin B per ml(growth medium). Confluent monolayers were maintained inEagle's MEM containing the same supplements except thatthe concentration of fetal calf serum was reduced to 1%(maintenance medium). HCMV strain AD169 is the GCV-sensitive wild-type virus. D1/3/4, D6/3/1, and D10/3/2 areGCV-resistant mutants derived from AD169. HP/AlB/4 is arecombinant resulting from transfer of the D1/3/4 DNApolymerase mutation to AD169.

Antiviral agents. GCV was obtained as a gift from Syntex,Palo Alto, Calif. HPMPA and HPMPC were gifts fromBristol-Myers Squibb, Wallingford, Conn. AraT and PFAwere purchased from Sigma, St. Louis, Mo. ACV waspurchased from Burroughs-Wellcome, Research TrianglePark, N.C.

Antiviral susceptibility measured by plaque reduction assay.All monolayers in a 24-well plate were inoculated with asingle dilution of virus calculated to produce 30 to 40 plaquesin the absence of drug. Four control wells were overlaid withmaintenance medium containing 0.3% agarose and no drug.The remaining wells were overlaid with medium containing0.3% agarose and a range of drug concentrations. Quadru-plicate wells were exposed to each drug concentration. The50% effective dose (ED50) was determined as the micromolarconcentration of each antiviral agent which reduced thenumber of plaques by 50% or greater relative to the numberof plaques produced in the control wells. The assay resultswere read at 10 to 14 days postinfection.

Antiviral susceptibility measured by ELISA. Monolayers in96-well microtiter plates were inoculated with wild-type ormutant strains of HCMV at a single dilution calculated toproduce 50% (2+) cytopathic effect in 5 to 7 days. A set ofsix cell control wells were mock infected with maintenancemedium alone. The inoculum was replaced with mediumcontaining dilutions of the drug (six wells per dilution).Medium without drug was added to the cell control wells aswell as one set of inoculated wells which served as the viruscontrol. HCMV late antigen was detected by enzyme-linkedimmunosorbent assay (ELISA) as described previously (33).The ED50 is the micromolar concentration of each antiviralagent which reduced the absorbance of colored substrateproduct by 50% or greater.HPLC analysis of GCV anabolism in virus-infected cells.

HFF monolayers in 150-cm2 flasks were inoculated at a

multiplicity of infection of 1.0. One flask was mock infectedto serve as an uninfected-cell control. At either 72 or 96 hpostinfection, 25 ,uCi of [3H]GCV (a gift from Syntex) wasadded to each flask.

After 24 h of incubation, the monolayers were trypsinized.Cells were washed three times with Dulbecco's phosphate-buffered saline (PBS). After the final wash, the cells were

resuspended in PBS and extracted with 0.5 M perchloricacid. The extracts were centrifuged, and the supernatant wascollected for high-pressure liquid chromatography (HPLC)analysis.GCV and GCV monophosphate (a gift from Syntex) at

concentrations of 100 mM were used as standards for HPLCanalysis of the products of infected-cell GCV anabolism.HPLC was performed on a Partisil 10 ODS3 reverse-phasecolumn (250 by 4.6 mm; Phenomenex, Rancho Palos Verdes,

Calif.) with 0.02 M KH2PO4 (pH 3.65) as the mobile phase.One-hundred-microliter samples of the perchloric acid ex-tracts were injected onto the column. Fractions were col-lected at 0.3-min intervals for 15 min. The radioactivity ofeach fraction was measured by scintillation counting.DNA sequencing. Plasmid clones of the HCMV DNA

polymerase genes were identified from genomic libraries ofeach of the three GCV-resistant mutants D1/3/4, D6/3/1, andD10/3/2, the recombinant HP/AlB/4, and the wild-typeAD169 strain from which the mutants were derived. Theseplasmids were purified on CsCl gradients. For sequencing, 5to 7 ,ug of double-stranded plasmid DNA was denatured in200 mM NaOH-0.2 mM EDTA at room temperature for 5min, neutralized with 200 mM ammonium acetate (pH 4.6),and ethanol precipitated. The denatured DNA was used asthe template for sequence analysis by the Sequenase version2.0 protocol (United States Biochemical, Cleveland, Ohio).

Transfections. Monolayers of HFF in 60-mm dishes wereinfected with wild-type AD169 virus at a multiplicity ofinfection of 0.5. The virus was allowed to adsorb for 3 h.Plasmid DNA (40 p,g) carrying the DNA polymerase genederived from a GCV-resistant mutant was mixed with 1 ml ofgrowth medium and 10 RI of hexadimethrine bromide (Sig-ma) (27). The DNA solution was placed on the infected cellsfor 6 h. The cells were washed with Earle's balanced saltsolution and treated with 30% dimethyl sulfoxide for 4 min.The dimethyl sulfoxide was removed; the cells were washedtwice with Earle's balanced salt solution, and growth me-dium was added. After 72 h, the remaining viable cells wereharvested and used to infect new HFF monolayers in six-well plates. HPMPA at a concentration of 1 ,uM was addedto the maintenance medium for selection of recombinantvirus. The supematant fluid was harvested after 14 days andused to inoculate HFF monolayers in 24-well plates. Themonolayers were overlaid with 0.3% agarose containingmaintenance medium and 1.5 FM HPMPA. Titers weredetermined in the presence and absence of drug. Isolatedplaques appearing in wells with drug were selected andpassed to tubes of HFF cells, with continued selection inmedium containing 1.5 ,uM HPMPA. Supernatant virus fromthe tube cultures was plaque purified again in the presence ofHPMPA. Virus samples from isolated plaques were thenpropagated without selection, plaque purified, and main-tained frozen at -70°C for further analysis.

RESULTS

Isolation of GCV-resistant mutants. The GCV-resistantmutants were isolated by exposure of the wild-type stock toincreasing concentrations of GCV. Ten tube monolayerswere inoculated with AD169 and maintained in mediumcontaining 5 ,uM GCV. Each tube monolayer was passedseparately and exposed to increasing concentrations of GCVin 10 ,uM increments up to 25 puM. Tube isolates whichshowed 100% cytopathic effect at 25 ,uM were tested at 50and 100 FM GCV. Three of these tube isolates showed 100%cytopathic effect in the presence of 100 p,M GCV. Each ofthese isolates was plaque purified three times. The plaque-purified strains have been designated D1/3/4, D6/3/1, andD10/3/2.

Initially, attempts were made to isolate GCV-resistantmutants from the wild-type AD169 stock by direct selectionin the presence of 5, 10, or 25 ,uM GCV. Samples from thefew plaques which appeared were retested for resistance toGCV. In all cases, no increased resistance relative to theparental wild-type AD169 could be detected even after

VOL. 66, 1992

on August 18, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

7148 LURAIN ET AL.

TABLE 1. Susceptibility testing results for HCMV strainsa

Assay and ED50 (PLM)virus GCV ACV PFA AraT HPMPA HPMPC

Plaque reductionassay

AD169 10 150 250 4,000 1.0 1.0D6/3/1 200 150 300 200 4 8D1/3/4 200 150 300 100 4 8D10/3/2 200 150 300 < 100 4 8HP/A1B/4 50 100 300 400 4 8

In situ ELISAAD169 6.25 200 200 4,000 1.0 1.0D6/3/1 200 100 200 200 5 10D1/3/4 200 150 200 200 5 10D10/3/2 200 150 150 100 10 15HP/A1B/4 50 100 300 200 5 15a Both the plaque reduction assay and the in situ ELISA were repeated two

or more times, and the data presented are derived from a representative assay.Sensitive and resistant strains were always assayed simultaneously forcomparison.

several rounds of plaque purification. Similar results havebeen reported by others (2, 31).

Antiviral susceptibility testing. The antiviral susceptibilityprofiles of mutants D1/3/4, D6/3/1, and D10/3/2 and thewild-type AD169 were determined by the plaque reductionassay (Table 1). These mutants were 10-fold more resistantto GCV than wild-type AD169. The mutants did not showincreased resistance to the pyrophosphate analog PFA or thenucleoside analog ACV. Unlike AD169, the mutants were,however, resistant to the nucleoside monophosphate analogsHPMPA and HPMPC and hypersensitive to AraT.An in situ ELISA (33) was used as an alternative method

to determine the susceptibility profiles of HCMV strains.This assay detects a late viral antigen produced in infectedfibroblast monolayers. Production of the antigen is depen-dent on DNA replication, which is inhibited by all of theseantiviral agents. The results are shown in Table 1. Theantiviral susceptibility profiles of mutant and wild-typestrains demonstrated by the plaque reduction assay wereconfirmed by the ELISA.

Analysis of GCV anabolism by HPLC. GCV anabolism inmutant- and wild-type-virus-infected cells was analyzed byreverse-phase HPLC. Standard retention times for unphos-phorylated GCV and GCV monophosphate averaged 10 and4 min, respectively. No standards are available for the di-and triphosphate derivatives of GCV; however, other nucle-otide di- and triphosphates had retention times of between 2and 3 min in this assay system. It was assumed that the di-and triphosphates of GCV would likewise have shorterretention times than the GCV monophosphate on this re-verse-phase column.

In uninfected fibroblasts (Fig. 1A), very low levels ofphosphoxylated GCV derivatives were present in the frac-tions with retention times that corresponded to those of thenucleoside phosphates (fractions 8 to 18). Unphosphorylateddrug eluted in fractions 33 to 38. The levels of phosphory-lated GCV derivatives in cells infected with wild-type virusare shown in Fig. 1C (fractions 8 to 18). The correspondingfractions from mutant-infected cells had 10-fold lower countsthan those from wild-type-infected cells (Fig. 1B, D, and F).GCV anabolism in mutant-infected cells is therefore greatlyreduced compared with that in wild-type-infected cells.These results suggest that each of the mutants has analteration in the kinase that phosphorylates GCV (22, 32).

DNA sequence of mutant DNA polymerase genes. Althoughthe HPLC data indicated that the GCV resistance of themutants could result from reduced GCV anabolism, theresistance to HPMPA and HPMPC suggested that there wasan additional mutation in the DNA polymerase gene of eachof these strains. The nucleotide sequence of the entire DNApolymerase gene from each of the three GCV-resistantmutants was therefore determined and compared with thepublished sequence for the wild-type AD169 (20). Eachmutant polymerase gene was found to contain a single basechange that results in an amino acid substitution (Fig. 2). Theposition numbers in Fig. 2 correspond to those of thepublished HCMV AD169 DNA polymerase sequence (20).The mutations in both D1/3/4 and D6/3/1 are at position 2160,where A is substituted for C. This mutation results in aconservative amino acid change from leucine to isoleucine atamino acid residue 501. The mutation in D10/3/2 is atposition 1893, where G is substituted for T, which changesamino acid residue 412 from phenylalanine to valine. Asecond, silent mutation in D10/3/2 at position 2111 (C to A)was also found. The DNA polymerase gene of the wild-typeAD169 strain from which the mutants were derived wassequenced wherever mutations were found in D1/3/4, D6/3/1,and D10/3/2 in order to eliminate the possibility of AD169strain-specific changes. The sequence of the strain of AD169used in this study is identical to the published AD169sequence (20) at every site at which a base substitution hasbeen found in D1/3/4, D6/3/1, and D10/3/2.Marker transfer of the D1/3/4 mutation to the wild-type

AD169 background. The HPLC and sequence data demon-strated that there are two mutations in the GCV-resistantstrains D1/3/4, D6/3/1, and D10/3/2. We therefore wanted todetermine the contribution of the mutant DNA polymeraseto GCV resistance. A plasmid carrying the D1/3/4 DNApolymerase gene was transfected into HFF infected with thewild-type strain AD169. Resistant virus was selected in thepresence of 1.5 ,uM HPMPA, and strain HP/AlB/4 wasobtained from this selection. The ED50 of GCV for HP/AlB/4 is 50 p,M, which represents a fourfold reductioncompared with that for the parental strain D1/3/4 (Table 1);however, HP/AlB/4 remains fivefold more resistant to GCVthan the wild-type AD169. The ED50 of HPMPA (4 ,uM) forHP/A1B/4 is similar to that for D1/3/4. HP/AlB/4, likeD1/3/4, is also hypersensitive to AraT and sensitive to PFA.Although HP/A1B/4 is resistant to GCV, the anabolism of

GCV in HP/AlB/4-infected cells is similar to that in wild-type AD169-infected cells (Fig. 1E). By comparison, GCVanabolism in cells infected with the parental D1/3/4 virus is atleast 10-fold lower than that in cells infected with the wildtype. These data indicate that the kinase that phosphorylatesGCV is unaltered in HP/AlB/4.Nucleotide sequence analysis of the HP/AlB/4 DNA

polymerase gene demonstrated that the same base substitu-tion at position 2160 (C to A) found in the parental mutantD1/3/4 is present in HP/AlB/4 (Fig. 3). HP/AlB/4, therefore,is a recombinant strain resulting from transfer of the D1/3/4polymerase mutation to the wild-type AD169.

Relationship of mutations to conserved regions of otherDNA polymerases. Comparison of the nucleotide sequencesof the mutant polymerase genes with the sequences of otherviral DNA polymerases shows a high degree of conservationin the regions where we have mapped mutations (Fig. 4). Theleucine residue at position 501 which is changed to isoleucinein both D6/3/1 and D1/3/4 is conserved in all human herpes-virus DNA polymerases (34). The phenylalanine residue atposition 412 which is changed to valine in D10/3/2 is con-

J. VIROL.

on August 18, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

POLYMERASE MUTANTS OF HUMAN CYTOMEGALOVIRUS 7149

1,500. A

2,O . ........................................

1,§ 0 ._ ................................................................................................................................................................................................................

1,a . ................................................................................................................................................................................................................

t_

. ................................................................................................................... ............................. .............................5 i15 25 5

2,500C

2Z 0 ..........................................................

1,5............................0..........................................................0..............

5 I5 25 3U

_..........................................................................................................................................................................................................

_ .......................................................................................................... ...........

_.................................................. ....................... .................................................................5f 15 25 5

D

. .............................................................................................................................................................................. .............................

..............................................................................................................................................................

4.,.4.*I

2,000

1.500

1,000

5 15 25 5 15 25

Fraction numberFIG. 1. HPLC analysis of virus-induced GCV phosphorylation. Fractions 33 to 38 contain unphosphorylated GCV, and fractions 8 to 18

contain the phosphorylated GCV derivatives. (A) Mock-infected human fibroblasts. (B) Dl/3/4-infected fibroblasts. (C) AD169-infectedfibroblasts. (D) D6/3/1-infected fibroblasts. (E) HP/AlB/4-infected fibroblasts. (F) D10/3/2-infected fibroblasts. Standards consisting of GCVand GCV monophosphate were run on the same column to determine relative retention times. The average retention time for GCV was 10min (fractions 34 to 36), and that of GCV monophosphate was 4.5 min (fractions 15 and 16). (E) HP/AlB/4 was assayed at a later time. Withthis sample, the unphosphorylated GCV is contained in fractions 29 to 33 and the phosphorylated derivatives are in fractions 8 to 19.Standards and AD169-infected cells were run as controls. All results were reproduced in a minimum of three separate experiments.

served not only in all human herpesviruses (Fig. 4) but alsoin several other prokaryotic and eukaryotic DNA polymer-ases (1, 37). These mutations are within 300 bp of each otherin the coding sequence of the HCMV polymerase gene.Wong et al. (37) identified six regions of homology among

a group of eukaryotic and prokaryotic DNA polymerases.These regions were labeled I through VI in order of decreas-ing degree of homology. An additional region displayinghomology among several viral DNA polymerases but notfound in human polymerase a was described by Gibbs and

._

ai)

C)

40

0

VOL. 66, 1992

5 le 3525

on August 18, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

7150 LURAIN ET AL.

AACATCAACTCTTTTGACTTGAAGTACATCCTCACGCGTCTCGAGTACCTGTATAA_---- ----

_- -_- -_- -_- -_- -_-

- -- -- -- -- ---G--- -- -- -- -- -- ---_-__-__-_--_--_-__-__-_--_--_-__-__-_ --_

GACTCGCAGCGCTTCTGCAAGTTGCCTACGGCGCAGGGCGGCCGTTTCTTTTTACA

1940 HCMV AD169OGGTG D10/3/2

HHV6----- EBV

HSV- 12000 HSV-2&CAGC VZV

405 T G Y N I N S F405 T G Y N I N S V361 T G Y N I N N F376 T G Y N V A N F463 T G Y N I I N F464 T G Y N I I N F444 T G Y N I V N F

*

DLKYILTRD LKY I LT RDLKYLLIRD W P Y I L D RD W P F L LA KD W P F V L T KD W A F I M E K

AD169Dl/3/4D6/3/1Dl0/3/2

AD169Dlp3/4D6/3/1D10/3/2

2060CCCGCCGTGGGTTTTAAGCGGCAGTACGCCGCCGCTTTTCCCTCGGCTTCTCACAACAAT

HCMV AD169D1/3/4D6/3/1

------------------------------------------------ HHV6EBVHSV- 1

2120 HSV-2CCGGCCAGCACGGCCGCCACCAAGGTGTATATTGCGGGTTCGGTGGTTATCGACATGTAC VZV

2180AD169 CCTGTATGCATGGCCAAGACTAACTCGCCCAACTATAAGCTCAACACTATGGCCGAGCTTD1/3-4---------------------------------------A--------------------D6/3/1 -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- - --A-- -- -- -- -- -- -- -- -----

D10//23--2---------------------------------------------_--_-__-__-_--_-__-__-_--_--_-__-__-__

FIG. 2. Nucleotide changes in the DNA polymerase sequencesof HCMV AD169 mutants. *, no change in amino acid. Numbers atthe right correspond to the nucleotide positions of the HCMVAD169 EcoRI-M fragment (20).

coworkers (12) and designated A. Recently, Hwang et al.reported a seventh region of homology (17). All of theseregions are listed in Table 2 in the order in which they appearin the DNA polymerases relative to the N terminus. Thelinear order which is conserved among the polymerases isIV-A-II-VI-III-I-VII-V (17, 37). The mutation in D10/3/2 lieswithin region IV. The mutation in D1/3/4 and D6/3/1 is foundbetween regions IV and A. The homology among herpesvi-rus DNA polymerases sequences remains high betweenthese two regions, although the sequence conservation is notfound outside this virus group (12, 34).

DISCUSSIONIn this report, we describe the isolation of three GCV-

resistant mutants of HCMV. This high-level GCV resistancehas been shown to be the combined effect of changes in twoviral functions: the DNA polymerase and the kinase required

T C G A

a"

-r

a"

NW _

-

_ ___sk go

__ok

2T C G A

U-~

vap

_ :-

-plow_

__l4m_m

c3 C

T C G A

- _p

_....

W-now_% _

GGTATC

AC

AAC

TC* -AGAATA

FIG. 3. Autoradiogram showing partial nucleotide sequence ofthe DNA polymerase gene from (panel 1) the wild-type AD169,(panel 2) the mutant D1/3/4, and (panel 3) the recombinant HP/AlB/4. The base change from C in strain AD169 to A in D1/3/4 andHP/A1B/4 at position 2160 is indicated.

493 K T N S P493 K T N S P493 K T N S P433 K I T A Q448 K L S L S532 K I K L S533 K V K L S513 K L K L S

NYKLNTMAELY

NYKINTMAELY

NYKINTMAELYN Y K L D T I A K I CD Y K L D T V A R H LSYKLNAVAEAV

SYKLNAVAEAV

S Y K L D S V A R E A

FIG. 4. Amino acid sequences in regions surrounding the site ofsubstitutions in the HCMV mutants D1/3/4, D6/3/1, and D10/3/2.Residues substituted in these mutants are marked with an asterisk(*). The homologous sequences of the other herpesviruses areshown for comparison. The amino acid residue numbers are fromTeo et al. (34). HHV6, human herpesvirus 6; EBV, Epstein-Barrvirus; VZV, varicella-zoster virus.

for GCV phosphorylation (2, 22, 32). Unlike HSV, fewdrug-resistant mutants of HCMV have been studied furtherto determine the mechanism of resistance, and no mutationshave been mapped to specific nucleotides of the DNApolymerase gene.

All three of our HCMV mutants demonstrate markedreductions in infected-cell anabolism ofGCV compared withthe wild-type AD169 (Fig. 1). Others have reported similardecreased GCV anabolism in GCV-resistant mutants fromboth laboratory HCMV strains (2) and clinical HCMV iso-lates (30). In cells infected with HSV, GCV is monophos-phorylated by the virus-encoded thymidine kinase (29);however, HCMV does not encode a thymidine kinase ho-molog (4). Instead, GCV appears to be phosphorylated by arecently identified protein kinase homolog encoded by theHCMV UL97 open reading frame (22, 32). We have identi-fied a point mutation in the UL97 gene of one of our mutants(unpublished data), and therefore, we postulate that thereduced GCV anabolism in cells infected with our mutants isthe result of changes in this enzyme.The initial evidence that the polymerase genes of the

GCV-resistant strains might also contain mutations was theobserved cross-resistance to the nucleoside monophosphate

TABLE 2. Conserved regions of the DNA polymerases ofHCMV and HSV'

HCMV residues HSV residues Conserved region

379-421 437-479 IV538-598 577-637 A696-742 694-736 II771-790 772-791 VI805-845 805-845 III905-919 881-896 I962-970 938-946 VII978-988 953-963 V

a The regions are listed in order from the N terminus of the polymerase.Regions I to VI are from Wong et al. (37); region A is from Gibbs et al. (12);region VII is from Hwang et al. (17). The amino acid substitution in D10/3/2 isat residue 412 in region IV; the amino acid substitutions in D1/3/4 and D6/3/1are at residue 501, which lies between regions IV and A.

AD169D1/3/4D6/3/11D10/Z32

AD169DlZ3/4D6/3l/D1013/2

J. VIROL.

on August 18, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

POLYMERASE MUTANTS OF HUMAN CYTOMEGALOVIRUS 7151

analogs HPMPA and HPMPC. Sequencing of the DNApolymerase genes of the three mutants revealed that eachencodes a polymerase containing a single nucleotide basechange which produces an amino acid substitution. Thesame base substitution resulting in a conservative amino acidchange from leucine to isoleucine at residue 501 is present inboth D1/3/4 and D6/3/1. Since these two strains were isolatedindependently from the same AD169 stock, the presence ofthe same mutation may reflect both the limited number offunctional mutations conferring GCV resistance and thefrequency with which these mutations occur in the wild-typepopulation. Similar results have been reported by Larder etal. (21) for a series ofDNA polymerase mutants of HSV-1, ofwhich three have identical point mutations.

In order to demonstrate that a conservative substitutioncould result in drug resistance, the DNA polymerase gene ofD1/3/4 was transferred to the wild-type background. Nucle-otide sequence analysis demonstrated that the same basesubstitution found in the parental strain D1/3/4 was presentin the recombinant HP/AlB/4 (Fig. 4). When cells wereinfected with HP/AlB/4, wild-type GCV anabolic activitywas observed (Fig. 1). The HPMPA and HPMPC resistanceof the recombinant remains the same as that of the parentalmutant, but GCV resistance is reduced by 75% comparedwith that of the parental strain. This is not surprising, sincethe high-level GCV resistance of D1/3/4 is apparently theresult of two mutations. Transfer of only the polymerasemutation would be expected to lead to a lower level of GCVresistance. The fact that the recombinant HP/AlB/4 remainshypersensitive to AraT suggests that the altered DNA poly-merase may play a role in this hypersensitivity as well. Itappears that even a conservative amino acid change canhave a significant effect on drug susceptibility.The base substitution found in D10/3/2 is 267 bases in the

5' direction from that of the mutation found in D1/3/4 andD6/3/1. This mutation results in an amino acid change fromphenylalanine to valine at residue 412 in the polypeptide.Marker transfer experiments are under way to confirm thatthis mutation also confers resistance to HPMPA, HPMPC,and GCV as well as hypersensitivity to AraT.Both phenylalanine 412 and leucine 501 are conserved in

the DNA polymerases of all the other human herpesvirusesfor which sequence data are available: HSV-1 and HSV-2,Epstein-Barr virus, varicella-zoster virus, and human her-pesvirus 6 (20, 34). Phenylalanine 412 is also conserved inother prokaryotic and eukaryotic DNA polymerases, includ-ing human DNA polymerase a, 4)29, vaccinia virus, andSaccharomyces cerevisiae (1, 3, 24, 37) (Fig. 4). Althoughthese amino acid residues are highly conserved among DNApolymerases, the viability of these three HCMV mutants isnot significantly affected by a single conservative substitu-tion.Of the large number of drug resistance mutations which

have been assigned to the polymerase locus ofHSV (5, 6, 12,13, 16, 17, 21, 36), the majority have been mapped within theC-terminal conserved regions, which encode amino acidresidues 690 to 1100 (12, 37). Specifically, these residuescomprise conserved regions II, III, and V described byWong et al. (37) and the recently identified region VII (17)(Table 2). The HCMV mutants that we have isolated havebase changes toward the 5' end of the polymerase gene. Thesubstitution in D10/3/2 lies in conserved region IV, while thesubstitution found in D6/3/1 and D1/3/4 lies between regionsIV and A.Conserved regions IV and A have been predicted to be

components of a possible 3'--*5' exonuclease domain (1, 3,

15, 37). The evidence for this functional domain is based onsite-directed mutagenesis studies of the DNA polymerase of+29 by Bernad and co-workers (1). Conservation of thesesequences among DNA polymerases supports the hypothe-sis that this region of the enzyme has 3'- 5' exonucleaseactivity. In addition, 3'--5' exonuclease function has beenfound to be tightly associated with the DNA polymerases ofEpstein-Barr virus (35), HSV (19), and HCMV (25).More recent evidence, however, suggests that it may not

be possible to define the functional domains of the HSVpolymerase at the level of the primary amino acid sequence.Knopf and Weisshart (19) were able to neutralize both the3'-+5' exonuclease and polymerizing functions of the HSVDNA polymerase by using antibodies specific for each of theN-terminal, central, and C-terminal peptide regions. Gibbsand coworkers (14) constructed HSV mutants by site-spe-cific mutagenesis in region IV. These were base substitutionsresulting in conservative amino acid changes. Two of themutants were nonviable. Thus, the polymerization functionof the enzyme was apparently inhibited by a mutation in theregion thought to have 3'-*5' exonuclease function.

Other evidence suggests that the amino acid residues thatcontribute to the nucleotide and pyrophosphate binding sitesare not confined to the C-terminal portion of the polypeptide,as has been postulated previously (13). Crumpacker et al. (6)described a GCV-resistant mutant of HSV-2 which wassensitive to ACV and PAA. Wang et al. (36) isolatedsuppressor mutants of HSV from a parental aphidicolin-resistant, PAA-hypersensitive strain. The suppressor muta-tions which conferred resistance to PAA were mapped to theN-terminal portion of the polymerase gene.Our data support the growing body of evidence that the

nucleotide-binding site of the HSV polymerase may bealtered by changes at sites other than the C-terminal regionof the polypeptide. Thus, the amino acid substitutions in thepolymerases of these HCMV mutants could be involveddirectly in nucleotide binding or indirectly in determinationof the tertiary structure of the binding site. It is also possiblethat the drug resistance of the mutants could result directlyfrom altered 3'--5' exonuclease function. Further analysis ofthe polymerases encoded by these mutants should contrib-ute to the determination of the functional domains of thisenzyme.

ACKNOWLEDGMENTS

Grant support was provided by Potts Foundation grant LU#3043-RCC/Potts.We thank Lois Spafford and Lynn Hilger for their technical

assistance.

REFERENCES1. Bernad, A., L. Blanco, J. M. Lazaro, G. Martin, and M. Salas.

1989. A conserved 3'->5' exonuclease active site in prokaryoticand eukaryotic DNA polymerases. Cell 59:219-228.

2. Biron, K. K., J. A. Fyfe, S. C. Stanat, L. K. Leslie, J. B. Sorrell,C. U. Lambe, and D. M. Coen. 1986. A human cytomegalovirusmutant resistant to the nucleoside analog 9.{[2-hydroxy-1-(hydroxymethyl)ethoxy]methyl}guanine (BW B759U) inducesreduced levels of BW B769U triphosphate. Proc. Natl. Acad.Sci. USA 83:8769-8773.

3. Blanco, L., A. Bernad, M. A. Basco, and M. Salas. 1991. Ageneral structure for DNA-dependent DNA polymerases. Gene100:27-38.

4. Chee, M. S., A. T. Bankder, S. Beck, R. Bohne, C. M. Brown, R.Cerny, T. Horsnell, C. A. Hutchinson m, T. Kouzarides, J. A.Martignetti, E. Preddie, S. C. Satchweli, P. Tomlinson, K. M.Weston, and B. G. Barrell. 1990. Analysis of the protein-coding

VOL. 66, 1992

on August 18, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

7152 LURAIN ET AL.

content of the sequence of human cytomegalovirus strainAD169. Curr. Top. Microbiol. Immunol. 154:125-169.

5. Coen, D. M., H. E. Fleming, Jr., L. K. Leslie, and M. J.Retondo. 1985. Sensitivity of arabinosyladenine-resistant mu-tants of herpes simplex virus to other antiviral drugs andmapping of drug hypersensitivity mutations to the DNA poly-merase locus. J. Virol. 53:477-488.

6. Crumpacker, C. S., P. N. Kowalsky, S. A. Oliver, L. E.Schnippter, and A. K. Field. 1984. Resistance of herpes simplexvirus to 9-{[2-hydroxy-1-(hydroxy-methyl)ethoxy]methyl}gua-nine: physical mapping of drug synergism within the viral DNApolymerase locus. Proc. Natl. Acad. Sci. USA 81:1556-1560.

7. D'Aquila, R. T., and W. C. Summers. 1987. Isolation andcharacterization of phosphonoacetic acid-resistant mutants ofhuman cytomegalovirus. J. Virol. 61:1291-1295.

8. Darby, G., B. A. Larder, and M. M. Inglis. 1986. Evidence thatthe "active centre" of the herpes simplex virus thymidinekinase involves an interaction between three distinct regions ofthe polypeptide. J. Gen. Virol. 67:753-758.

9. Drew, L. W. 1988. Cytomegalovirus infection in patients withAIDS. J. Infect. Dis. 158:449-456.

10. Erice, A., S. Chou, K. K. Biron, S. C. Stanat, H. H. Balfour, andM. C. Jordan. 1989. Progressive disease due to ganciclovir-resistant cytomegalovirus in immunocompromised patients. N.Engl. J. Med. 320:289-292.

11. Forbes, B. A. 1989. Acquisition of cytomegalovirus infection: anupdate. Clin. Microbiol. Rev. 2:204-216.

12. Gibbs, J. S., H. C. Chiou, K. F. Bastow, Y.-C. Cheng, and D. M.Coen. 1988. Identification of amino acids in herpes simplex virusDNA polymerase involved in substrate and drug recognition.Proc. Natl. Acad. Sci. USA 85:6672-6676.

13. Gibbs, J. S., H. C. Chiou, J. D. Hall, D. W. Mount, M. J.Retondo, S. K. Weller, and D. M. Coen. 1985. Sequence andmapping analyses of the herpes simplex virus DNA polymerasegene predict a C-terminal substrate binding domain. Proc. Natl.Acad. Sci. USA 82:7969-7973.

14. Gibbs, J. S., K. Weisshart, P. Digard, A. de Bruynkops, D. M.Knipe, and D. M. Coen. 1991. Polymerization activity of ana-like DNA polymerase requires a conserved 3'-5' exonucleaseactive site. Mol. Cell. Biol. 11:4786-4795.

15. Haffey, M. L., J. Novotny, R. E. Bruccoleri, R. D. Carroll, J. T.Stevens, and J. T. Matthews. 1990. Structure-function studies ofthe herpes simplex virus type 1 DNA polymerase. J. Virol.64:5008-5018.

16. Honess, R. W., D. J. M. Purifoy, D. Young, R. Gopal, N.Cammack, and P. O'Hare. 1984. Single mutations at many siteswithin the DNA polymerase locus of herpes simplex viruses canconfer hypersensitivity to aphidicolin and resistance to phos-phonoacetic acid. J. Gen. Virol. 65:1-17.

17. Hwang, C. B. C., K. L. Ruffner, and D. M. Coen. 1992. A pointmutation within a distinct conserved region of the herpessimplex virus DNA polymerase gene confers drug resistance. J.Virol. 66:1774-1776.

18. Jacobson, M. A., W. L. Drew, J. Feinberg, J. J. O'Donnell, P. V.Whitmore, R. D. Miner, and D. Parenti. 1991. Foscarnet therapyfor ganciclovir-resistant cytomegalovirus retinitis in patientswith AIDS. J. Infect. Dis. 163:1348-1351.

19. Knopf, C. W., and K. Weisshart. 1990. Comparison of exonu-cleolytic activities of herpes simplex virus type-i DNA poly-merase DNase. Eur. J. Biochem. 191:263-273.

20. Kouzarides, T., A. T. Bankier, S. C. Satchwell, K. Weston, P.Tomlinson, and B. G. Barrell. 1987. Sequence and transcriptionanalysis of the human cytomegalovirus DNA polymerase gene.

J. Virol. 61:125-133.21. Larder, B. A., S. D. Kemp, and G. Darby. 1987. Related functional

domains in virus DNA polymerases. EMBO J. 6:169-175.22. Littler, D., A. D. Stuart, and M. S. Chee. 1992. Human cyto-

megalovirus UL97 open reading frame encodes a protein thatphosphorylates the antiviral nucleoside analogue ganciclovir.Nature (London) 358:160-162.

23. Miller, R. L., J. P. Utis, and F. Rapp. 1977. Differential effect ofarabinofuranosylthymine on the replication of human herpesvi-ruses. J. Virol. 23:679-684.

24. Morrison, A., J. B. Bell, T. A. Kunkel, and A. Sugino. 1991.Eukaryotic DNA polymerase amino acid sequence required for3'--5' exonuclease activity. Proc. Natl. Acad. Sci. USA 88:9473-9477.

25. Nishiyama, Y., K. Maeno, and S. Yoshida. 1983. Characteriza-tion of human cytomegalovirus-induced DNA polymerase andthe associated 3'--5' exonuclease. Virology 124:221-231.

26. Oberg, B. 1989. Antiviral effects of phosphonoformate (PFA,foscarnet sodium). Pharmacol. Ther. 40:213-285.

27. Rhim, J. S., J. B. Park, and G. Jay. 1989. Neoplastic transfor-mation of human keratinocytes by Polybrene-induced DNA-mediated transfer of an activated oncogene. Oncogene 4:1403-1409.

28. Shigeta, S., K. Konno, M. Baba, T. Yokota, and E. De Clercq.1991. Comparative inhibitory effects of nucleoside analogues ondifferent clinical isolates of human cytomegalovirus in vitro. J.Infect. Dis. 163:270-275.

29. Smee, D. F., R. Boehme, M. Chernow, B. P. Binko, and T. R.Matthews. 1985. Intracellular metabolism and enzymatic phos-phorylation of 9-(1,3-dihydroxy-2-propoxymethyl)guanine andacyclovir in herpes simplex virus-infected and uninfected cells.Biochem. Pharmacol. 34:1049-1056.

30. Stanat, S. C., J. E. Reardon, A. Erice, M. C. Jordan, W. L.Drew, and K. K. Biron. 1991. Ganciclovir-resistant cytomega-lovirus clinical isolates: mode of resistance to ganciclovir.Antimicrob. Agents Chemother. 35:2191-2197.

31. Sullivan, V., and D. M. Coen. 1991. Isolation of foscarnetvresistant human cytomegalovirus patterns of resistance andsensitivity to other antiviral drugs. J. Infect. Dis. 164:781-784.

32. Sullivan, V., C. L. Talarico, S. C. Stanat, M. Davis, D. M. Coen,and K. K. Biron. 1992. A protein kinase homologue controlsphosphorylation of ganciclovir in human cytomegalovirus-in-fected cells. Nature (London) 358:162-164.

33. Tatarowicz, W. A., N. S. Lurain, and K. D. Thompson. 1991. Insitu ELISA for the evaluation of antiviral compounds effectiveagainst human cytomegalovirus. J. Virol. Methods 35:207-215.

34. Teo, I. A., B. E. Griffin, and M. D. Jones. 1991. Characterizationof the DNA polymerase gene of human herpesvirus 6. J. Virol.65:4670-4680.

35. Tsurumi, T. 1991. Characterization of 3'-to-5' exonucleaseactivity associated with Epstein-Barr virus DNA polymerase.Virology 182:376-381.

36. Wang, Y., S. Woodward, and J. D. Hall. 1992. Use of suppres-sor analysis to identify DNA polymerase mutations in herpessimplex virus which affect deoxynucleoside triphosphate sub-strate specificity. J. Virol. 66:1814-1816.

37. Wong, S. W., A. F. Wahl, P.-M. Yuan, N. Arai, B. E. Pearson,K. Arai, D. Korn, J. W. Hunkapiller, and T. W.-F. Wang. 1988.Human DNA polymerase a gene expression is cell proliferationdependent and its primary structure is similar to both prokary-otic and eukaryotic replicative DNA polymerases. EMBO J.7:37-47.

J. VIROL.

on August 18, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from